Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add mayank-io/mstack/plugin install devWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/mayank-io/mstack/safe-pause)<a href="https://agentmods.dev/skills/mayank-io/mstack/safe-pause"><img src="https://agentmods.dev/badge/skills/mayank-io/mstack/safe-pause/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/mayank-io/mstack/safe-pause"><img src="https://agentmods.dev/badge/skills/mayank-io/mstack/safe-pause.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00075 | $0.01586 |
| Opus 5.5 | $0.00030 | $0.00634 |
| Sonnet 5.5 | $0.00015 | $0.00317 |
| Haiku 4.5 | $0.00007 | $0.00159 |
Grade A, and why
safe-pause scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 19d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 60 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Safe Pause
Bring this session and every subagent it owns to a stop that /dev:safe-resume can continue from without asking what was happening.
Violating the letter of these rules is violating their spirit. A pause that leaves one worktree dirty, pushes one branch, or forgets one owed step is not a pause.
Arguments
/dev:safe-pause [task|milestone|project] [--now]. Default task.
| Mode | Stop point | While running down |
|---|---|---|
task |
Each live subagent finishes the unit it is on (the commit, the test run, the review pass), commits, writes its resume memory, exits. Bound ~20 min. --now: stop at the last green point instead of finishing the unit. |
Session exits once the last subagent has. |
milestone |
Agents run to the end of the current milestone as its plan defines it: all its commits, its targeted tests, its review, READY reported. The next milestone is not started. | Session stays up to orchestrate, declines anything new, prints a progress line every 30 min. |
project |
The session completes the whole dispatched item: every milestone through LAND READY, and LANDED where a lander session exists. | Nothing new accepted during the run-down. |
Invariants (every mode)
- No new dispatch or backlog item is taken from the moment the skill starts.
- Peer messages that arrive are answered
PAUSED (<mode>), resumes on MK's wordand logged into RESUME.md under Peer messages logged. - Every worktree this session or its agents own ends with a commit. If the work is not a real commit, make a labelled one:
wip(pause): <what is built> / <what is owed>. - Nothing pushed, merged or landed. Never touch a primary main.
- One-full-suite rule: do not start a full suite to "finish". A suite already running may complete.
- Resume memory goes to
~/.local/state/mk-pause/<session-name>/RESUME.md. Never/tmp.
Procedure
Use bash with set -euo pipefail and "${VAR:?}" for every path variable. PS below is python3 ${CLAUDE_PLUGIN_ROOT}/scripts/pause_state.py (run --help for the subcommands).
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 19d ago First seen · 60 lines · 75 tokens per session scan A 46b132ef6b80
safe-pause is a skill published in the GitHub repository mayank-io/mstack (6 stars, last pushed 3d ago), licensed MIT. It adds 75 tokens to every session and 1,586 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-13.
Other skills, from other repositories
Dotfiles
Manage /.zshrc and a chosen set of /.config/ directories (you name them in Config.md) against a private GitHub dotfiles repo. Intents: Push (verify private remote, scan secrets), Pull (clone/diff), Sync, Restore (timestamped backup), Status, Backup (local timestamped copy). Security scan blocks .env, secrets, tokens…
conventional-commit
Rules for writing Conventional Commits messages in this project. MUST be invoked every time a commit is created.
python-release
Handle Python SDK release, build, bump, packaging metadata, PyPI client pin, uv.lock, nox/build workflow, and publish verification changes. Use for Python release process work or dependency pin bumps; do not use for ordinary Python feature implementation.
release-notes
Generate and publish concise, evidence-based notes in the body of the latest existing GitHub Release. Use only when the user explicitly invokes $release-notes or explicitly asks to update the latest existing GitHub Release body. Do not invoke for general release planning, changelog, tag, or version tasks.
ratchet-files
Ratchet source file size so every file stays small enough for an agent to read whole: find every file over a token budget, split oversized files by concept, and set up the pre-commit ratchet that stops them from growing. Use when a commit is blocked because a file is over the size budget, when asked to find or split…
git-guardrails-claude-code
A Claude Code setup that blocks selected destructive Git commands before they run. Git is a tool for tracking code changes; the blocked commands include pushing, deleting branches, discarding changes, and hard resets.